Blog
Bridging commentary on the art & science of content; content marketing, branded content, native advertising and the analytics, attribution and intelligence that powers it.
First load up the article, pull up your content analytics, then start to look through the data. Does the article perform above benchmarks? Below? Are people reading all of it? Is there a common drop off point? What devices are they reading on? Was it at expectations / above?… Continue reading
Interrogate the data, how was this data created? Is it clean? Does the recording of the data reflect the intention for how you want to use the data. Does it reflect the right date ranges? See if you can find some comparable data, or benchmarks from older data. To help contextualize your data. Continue reading
The first watch out in understanding a chart is looking at the scales. These can be easily distorted to misrepresent findings. Take a chart that is a portrait and make it landscape and it instantly stretches out. So do compare the scales, and the metrics used on each axis. Then explore the data itself, how was it collected, how many data points are in here. Can you draw or add any trendiness? Is the data substantial enough… Continue reading
Data really really helps in analyzing a video. So first stop is to go and get all your data on the video. Then re-watch the video, comparing how you watch it versus the data you have on its performance. Stop where people often stop, click where people are clicking. This will help contextualize and understand what the data is telling you. It may be obvious why people drop off at a certain point. Then you can go… Continue reading
Load up all the data you have on an ad. How many people saw it, what was the click-through rate, what was the view ability or attention on it? Then compare that against your average or your benchmark. Is that good, is that bad? Did the ad drive conversions? More than you expected or less? Answering these questions will help understand whether it is working or not. Continue reading
The scope of this could vary dramatically but the basics of a data analysis report include An executive summary, what is that the data is showing us and what did we find. Methodology, how was the data collected Constraints/limitations of the analysis/data, establishing boundaries around the data, how it can be used, what are the strengths and limitations of it. Benchmarks, benchmarks are helpful for contextualizing performance and nuance. Continue reading
When critically analyzing a report, make sure you read it start to finish. Often a nice trick is to start at the back and read all the addendums before you read the actual report. This is often where the limitations or constraints are placed. And knowing these before you interpret the report is helpful. Print it out, get a cup of coffee, get a pen, go to a quiet place. Read it through. Write notes. What surprised you,… Continue reading
A/B Testing is a tremendous practice for marketers to test two different variations. However, it is not bulletproof. First, you need to think about context, what happened before/after people engaged with this A/B test. You want to make sure you havenโt tipped the results one way with a prior A/B test earlier in the consumer experience. Then you need to establish is this a meaningful sample size. For that, you want to look at your overall population of… Continue reading
In analyzing an audience, you must understand how the audience data was collected. Is this first-party observed data, merged data, sampled data. How are the demographics for the data collected? Establishing these helps you understand your data foundation, is what youโve got rock-solid enough to build conclusions on top of. If it isnโt, where is it weak, as you need to acknowledge that in your conclusions. Often we have to make decisions off incomplete or proxy data. But… Continue reading
First, you must interrogate and understand how the demographic data was collected. This can provide limitations, nuance, and context which is important. For example, if it is inferred data, you may interpret your outcomes differently. Then break apart your dataset into meaningful demographic clusters, i.e. 18-24. And analyze your data that way. How does each demographic differ from the rest? Are there trends or patterns. Can you predict what a demographic will do? Are you beginning to establish… Continue reading